A guide of guides for industrial focus.
Brief start about Legends and Axes
and Titles
Figure 1. A widget example using titles, axes and legends
According to Mackinlay Card in “Information Visualization : Using Vision to think “, [1] visualization is to build up“ visual representations of abstract data to amplify cognition.” But, we still need to explicitly point out to what they correspond to.
That is where axes, legends and titles intervene.
As Leland Wilkinson states in “The grammar of Graphics”, [2] they give us absolute directions in a relative world “.
History of Legends and Axes and Titles,
Let’s get a few centuries back, actually let’s go straight to 123 B.C in which practice of marking distance was initiated by the Roman Tribune 123.BC with milestones.
According to Max Lay in “ The History of Traffic signs “ [3] as traffic increased, signing grew in importance. In fact, the more we have data circulating, the more we tend to lose focus on what's happening.
That is where guides, axes and legends become crucial: to stay focused and extend our cognitive capacities.
Personally, communication and road construction systems during the Roman Empire impress me as much as Dashboards and Reports !
Figure 2. Milario ( Milestone ) from the roman road that ran through Boeotia,
2nd BC
Now that we took the dust off of our subject, let’s present how to apply legends, titles and axes to our visual representations.
Applying Legends And Axes
Notions
First things first, let’s start with our notions.
Legend : A legend is a frame containing a set of elements illustrating how to identify a data in a visual representation.
Axe: An axe is a line providing context to the visual marks of a given chart. A chart without axes might be meaningful, but axes without visual marks are always irrelevant. So axes are always optional but useful elements for understanding the content of a chart.
Title: A title is a visual field containing a set of alphanumeric characters. As we saw for the Axe definition, titles provide context but in a much more precise, lightweight way to our audience. A visualization designer uses titles for individual charts, but as well for reports or dashboards.
Let’s Apply : from Bad to Worse.
In this section, we will provide some examples of applying legends, axes and titles to some charts and provide some design tips.
Titles
Figure 3. A title implementation example
using a circular gauge
Figure 3 shows us a circular gauge chart example. We have a title. This is a bad design example.
Figure 4. A title implementation example no 2
using a circular gauge
Figure 4 shows us another example using the same chart. This one is even worse than the previous one.
In fact, in figure 3, our title indicates to us that it is a gauge which is correct. Yet the title doesn’t ease our understanding of what really is our chart about.
Looking at figure 4, it partially tries to help us understand what our char is about. But what makes it worse than the previous one is that :
It is too descriptive and long for the available visual representation: It does not provide more context than we see.
It is uncertain whether red represents good or bad. We don’t have a legend available, so we should either add one, or remove that part from the title.
It is difficult to notice we have a title because of its poor formatting.
Quick tips:
Considering our examples, we can give the following quick fixes:
Keep the title as simple but as complete as possible. In order to do this, summarize what we see in two sentences, then one. Finally, repeat the same exercise, summarizing what we see to someone unaware of the subject.
Be sure that our title doesn’t mislead our audience; We should associate keywords of our title with the rest of our visualizations component. For example, in figure 4, I checked how we know if something is “bad” without any answer found.
Axes
Figure 5. An axe implementation example using a line chart
Figure 5 presents us with a line chart. We can see that the axes provide some context like country names and values, but we don’t know to what they correspond to.
Figure 6. An axe implementation example using a line chart no 2
This line chart in figure 6 is an example visually clean but misleading and dangerous. In fact, our horizontal axis represents categorical context (countries). Our vertical axis shows off values. But the choice of line marks, and our “countries” axis don’t fit together. It induces the following message “ my country's size evolves between itself “ which is absurd. Also, our “size” axis uses a SUM aggregation, but its title lacks precision.
Quick tips:
Verify if your visual marks (lines, areas, symbols etc) are coherent with your axis choices.
Use clean and precise axis titles. We can check the same exercise as for title, but reducing word count instead of sentence.
Legends
Figure 7. A legend implementation example using a choropleth map
Figure 7 shows us a choropleth map. We can see some beautiful colours, but actually we don’t know their meaning. We are missing our legend !
Figure 8. A legend implementation example using a choropleth map no 2
This example, in figure 8, tries to correct the previous one by adding legends and a legend title. But this example presents several problems: It shows too many elements, and the legend is oversaturated. We could keep in mind what Miller George.A stated in “The magical number seven, plus or minus two: Some limits on our capacity for processing information “ that [4] below seven the subjects were said to subitize.
Also, our colour palette tends to categorize our visual data yet the elements sub-description ( 2-3 , 3-4) follows a sequential order.
Quick tips:
Try to limit legend elements size to 7.
Always map visual entities with at least one coherent legend in our dashboards.
A result of our playground
Based on the self critics we made to ourselves, let’s provide an improved version for the 3 types of examples we saw in the previous sections:
Figure 9. An improved legend implementation example using a circular gauge chart.
Figure 10. An improved axe implementation example using a bar chart instead of a line chart.
Figure 11. An improved legend implementation using a choropleth map.
One step further
This section will provide some non-essential yet powerful options that today's visualization tools offer.
Grid Lines
Grid lines are useful elements that draw our attention to particular areas in our charts.
Figure 12. A grid line implementation example using a bar chart.
Tooltips
We can use tooltips, powerful interactive tools to annotate, emphasize the usage of our legends
Figure 13. A tooltip implementation example using a choropleth map.
Scaling
Scaling our axis based on our data repartition helps us compare the overall content.
Figure 14. A linear scaling implementation example using a bar chart .
Figure 15. A logarithmic scaling implementation example using a bar map.
Two steps further: The grammar of graphics
Before concluding, we should keep in mind that charts don't define titles, axes, and legends, but rather the inverse. A chart is actually built with different parts such as titles, axes, legends, visual marks and scaling. Charting is more intuitive, graphics theory is hidden and beautiful.
Also, if we want to rely on scalable, creative and impactful visual representations, we should listen to Leland Wilkinson who stated in " The Grammar of Graphics " [5] Elegant design requires us to think about a theory of graphics, not charts".
Conclusion
We can conclude by pointing out once more that titles, legends and axes are important players of our visualization process. They describe us context and associate visual marks with the latter. They also help us check the coherence between our visual entities and our abstract data.
Reference
[1] Card, S.K., Mackinlay, J.D. and Shneiderman, B. (2007). Readings in information visualization : using vision to think. San Francisco, Calif.: Morgan Kaufmann.
[2] Wilkinson, L. (2012). Grammar of graphics. Springer.
[3] Lay, M. G. (1993) ‘The history of traffic signs’, στο.
[4] Miller, G.A. (1994). The magical number seven, plus or minus two : some limits on our capacity for processing information. Washington, D.C.: American Psychological Assoc.
[5] Wilkinson, L. (2012). Grammar of graphics. Springer.
Comments